Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Building a hierarchy with neural networks: an example-image vector quantization.

L D Jackel, R E Howard, J S Denker

    Applied Optics
    |June 5, 2010
    PubMed
    Summary
    This summary is machine-generated.

    Related Concept Videos

    Gradient Vectors and Their Applications01:19

    Gradient Vectors and Their Applications

    Every point on a topographical map corresponds to a particular elevation, so the landscape can be modeled as a surface whose height depends on horizontal position. From any given location, a hiker may face infinitely many directions, but only one direction produces the fastest possible increase in elevation. This unique route is called the direction of steepest ascent, and in multivariable calculus, it is represented by the gradient vector of the elevation function.The gradient vector points...

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Measurement and implications of Saturn's gravity field and ring mass.

    Science (New York, N.Y.)·2019
    Same author

    Programme evaluation: What we can learn from the United States experience.

    The Australian journal of physiotherapy·2014
    Same author

    Graphene as a subnanometre trans-electrode membrane.

    Nature·2010
    Same author

    Electronic neural network chips.

    Applied optics·2010
    Same author

    Evolutionary expansion and specialization of the PDZ domains.

    Molecular biology and evolution·2009
    Same author

    Amplification in the auditory periphery: the effect of coupling tuning mechanisms.

    Physical review. E, Statistical, nonlinear, and soft matter physics·2007
    Same journal

    Multifunctional reconfigurable terahertz metasurface based on vanadium dioxide phase transition: achieving broadband absorption and efficient polarization conversion.

    Applied optics·2026
    Same journal

    High-Q-factor electromagnetically induced transparency utilizing quasi-bound states in the continuum in an all-dielectric terahertz metasurface.

    Applied optics·2026
    Same journal

    Automated stitching interferometry for high-precision metrology of X-ray mirrors.

    Applied optics·2026
    Same journal

    Experimental demonstration of an approach to designing a metal-dielectric DBR resonant cavity structure.

    Applied optics·2026
    Same journal

    High-precision wavefront reconstruction from a single-shot interferogram using a physics-driven hybrid feature calibration network.

    Applied optics·2026
    Same journal

    Ultra-high-Q Fano resonance based on coupled topological corner states in Kagome photonic crystals.

    Applied optics·2026
    See all related articles

    Electronic neural networks offer associative memory, but storage capacity is limited. This study introduces a hierarchical network approach to efficiently search vast memory lists beyond single network capabilities, demonstrated via image vector quantization.

    Area of Science:

    • Artificial Intelligence
    • Computational Neuroscience
    • Machine Learning

    Background:

    • Electronic neural networks function as associative memories, performing content-addressable search and error correction.
    • The memory storage capacity of single electronic neural networks is constrained by the number of neurons.
    • Efficiently searching large, stored memory datasets remains a challenge for conventional neural network architectures.

    Purpose of the Study:

    • To propose a novel method for enhancing the memory capacity of electronic neural networks.
    • To enable fast, parallel searching through memory lists exceeding the capacity of a single network.
    • To demonstrate the efficacy of a hierarchical network approach in associative memory tasks.

    Main Methods:

    • Development of a hierarchical network architecture comprising multiple interconnected neural networks.

    Related Experiment Videos

  • Implementation of a fast parallel search algorithm optimized for the hierarchical structure.
  • Application and validation of the proposed method using image vector quantization as a case study.

    • Main Results:

    • The hierarchical network successfully enabled searching through memory lists significantly larger than what a single network could store.
    • The parallel search mechanism demonstrated high efficiency in retrieving relevant memories.
    • Image vector quantization using the hierarchical approach yielded comparable or improved results to single-network methods.

    Conclusions:

    • Hierarchical neural network architectures offer a scalable solution for associative memory with expanded storage capacity.
    • This approach overcomes the limitations of single networks, facilitating efficient content-addressable search in large datasets.
    • The demonstrated success in image vector quantization suggests broad applicability in various pattern recognition and data retrieval tasks.